Synthesis, cure kinetics, and physical properties of a new tricyanate ester with enhanced molecular flexibility

Article abstract of DOI:10.1016/j.polymer.2011.07.024

1,2,3-Tris(4-cyanatophenyl)propane, a new tricyanate ester monomer that was designed to incorporate more flexible chemical linkages at junction points in the cured macromolecular network, was synthesized in nine steps with an overall yield of 26%. The highly purified monomer exhibited an activation energy of 110 kJ/mol for auto-catalytic cure at temperatures of 210 °C-290 °C, modestly lower than the comparably measured activation energy of a commercial cyanated novolac. The overall extent of cure achievable at these temperatures was also higher for the new monomer. Many physical properties of the cured monomer, including density, thermochemical stability, moisture uptake, and the impact of hydrolytic degradation on glass transition temperature were similar to those of commercial tricyanates, with a dry glass transition temperature at full conversion of at least 340 °C. These results illustrate how careful control of the local chemical structure in the vicinity of network junction points may be utilized to improve the properties of thermosetting polymer networks.

Full text of DOI:10.1016/j.polymer.2011.07.024

Polymer 52 (2011) 3933e3942
Contents lists available at ScienceDirect
Polymer
journal homepage: www.elsevier.com/locate/polymer
Synthesis, cure kinetics, and physical properties of a new tricyanate ester
with enhanced molecular flexibility
,
b
c
a
b
Andrew J. Guenthner a , Matthew C. Davis , Kevin R. Lamison , Gregory R. Yandek , Lee R. Cambrea ,
*
Thomas J. Groshens b, Lawrence C. Baldwin b, Joseph M. Mabry a
a Air Force Research Laboratory, Edwards AFB, CA 93524, USA
b Naval Air Warfare Center, Weapons Division, China Lake, CA 93555, USA
c ERC Corporation, Edwards AFB, CA 93524, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
1,2,3-Tris(4-cyanatophenyl)propane, a new tricyanate ester monomer that was designed to incorporate
more flexible chemical linkages at junction points in the cured macromolecular network, was synthe-
sized in nine steps with an overall yield of 26%. The highly purified monomer exhibited an activation
energy of 110 kJ/mol for auto-catalytic cure at temperatures of 210 ꢀCe290 ꢀC, modestly lower than the
comparably measured activation energy of a commercial cyanated novolac. The overall extent of cure
achievable at these temperatures was also higher for the new monomer. Many physical properties of the
cured monomer, including density, thermochemical stability, moisture uptake, and the impact of
hydrolytic degradation on glass transition temperature were similar to those of commercial tricyanates,
with a dry glass transition temperature at full conversion of at least 340 ꢀC. These results illustrate how
careful control of the local chemical structure in the vicinity of network junction points may be utilized
to improve the properties of thermosetting polymer networks.
Received 27 April 2011
Received in revised form
30 June 2011
Accepted 6 July 2011
Available online 23 July 2011
Keywords:
Cyanate ester
Cure kinetics
Activation energy
Published by Elsevier Ltd.
1. Introduction
The development of structureeproperty relationships for tri-
cyanate esters involves significant challenges not encountered in
In recent years, the aromatic tricyanate ester [1e3] has emerged
as an important class of high-performance thermosetting polymer,
combining high glass transition temperatures and short-term
thermal stability well in excess of 300 ꢀC with low-cost options for
composite processing such as filament winding and resin transfer
molding [4e6]. As a result, tricyanate esters have been used in
a growing array of applications in the aerospace industry [7,8],
where their high stiffness, low volatility, reduced toxicity, minimal
shrinkage during cure, and excellent resistance to fire and water
make them extremely well-suited to the fabrication of parts with
thick cross-sections. In addition, the relatively well-defined chemical
structure of networks formed from cyanate esters makes them ideal
candidates for the study of structureeproperty relationships in
thermosetting resins. Despite their many practical advantages,
however, tricyanate esters have received much less attention
recently than bisphenol type dicyanate esters [9] in terms of basic
research and the development of structureeproperty relationships.
studying most thermosetting resins. When fully cured, tricyanate
esters such as the phenyl triazine (“PT”, or “novolac-type”) resins,
exhibit network junction densities approximately double those of
the more common bisphenol type cyanate esters, due to the fact
that the central branch point (herein referred to as the “core”) in
the chemical structure of the monomer, forms a junction point for
polymer network segments, just as the triazine rings formed by
cyclotrimerization do (see Fig. 1 for an illustration). The high
density of network junctions constrains segmental motion, thereby
raising the glass transition temperature of fully trimerized systems
near to, and in some cases, beyond, the limit of even short-term
thermal stability of the molecules. The highly constrained motion
of network segments, however, makes full trimerization very
difficult to achieve in practice. Not surprisingly, then, the reported
glass transition temperatures of “cured” polycyanate esters varies
widely. In the case of cyanated novolac with the approximate
structure exemplified by PrimasetÒ PT-30, a commercial novolac
cyanate ester supplied by Lonza, shown in Fig. 1a, the reported dry
glass transition temperature of cured resin varies from 305 ꢀC [10]
to over 400 ꢀC [11]. In addition, the highly rigid chemical structure
of the network leads to a high level of brittleness [12], which
* Corresponding author. Tel.: þ1 661 275 8020.
E-mail addresses: andrew.guenthner@edwards.af.mil, andrewguenthner@gmail.
com (A.J. Guenthner).
0032-3861/$ e see front matter Published by Elsevier Ltd.
doi:10.1016/j.polymer.2011.07.024

3934
A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
Fig. 1. a) Chemical structure of monomer 9 compared with b) approximate chemical structure (n ¼ 1e2) for PrimasetÒ PT-30 resin c) after cure, the idealized network architecture
of 9 features a “checkerboard” pattern of relatively flexible trisubstituted propyl junctions (dashed circles) alternating with more rigid triazine ring junctions (dashed rounded
rectangles), d) whereas in PT-30 (shown with n ¼ 1 for convenience), the network architecture features only rigid triazine ring or phenyl ring junctions.
hinders fabrication and evaluation of test specimens for mechanical
property assessments. These challenges are not only an issue for
research investigations, but also represent obstacles to the wide-
spread use of these materials in large-scale fabrication.
One promising approach to overcoming the aforementioned
challenges is to synthesize new tricyanate esters with a broader
range of chemical structures and, hence, physical properties. By
comparing these new materials with already existing forms,
structureeproperty relationships can be more easily discerned.
Though derived specifically from data on cyanate ester networks,
these relationships are quite often applicable to thermoset
networks in general. At the same time, new monomers that exhibit

A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
3935
better cure and toughness characteristics can be used either in
place of or as co-monomer additives for existing commercial
products, making tricyanate esters a more attractive choice for
high-performance applications. Recently, for instance, Yameen
et al. [13] synthesized a new tricyanate ester oligomer that
combined excellent flow and processing characteristics with a glass
transition temperature of 280 ꢀC, equivalent to the best-performing
bisphenol type cyanate esters. Similarly, Cambrea et al. [14]
recently synthesized a tricyanate ester monomer that flowed at
room temperature and produced a dry glass transition temperature
of 305 ꢀC. In both these cases, a rigid phenyl ring was retained at the
monomer core, while a lower cross-link density compared to
products such as PrimasetÒ PT-30 led to somewhat lower glass
transition temperatures.
In the present work, an alternative approach to improving the
performance of tricyanate esters, that is also applicable to rigid
macromolecular networks in general, is explored. A new tricyanate
ester, designed to have the same cross-link density as PrimasetÒ PT-
30, yet with a more flexible core based on a 1,2,3-trisubstituted
propane, and with the chemical structure shown in Fig. 1b, was
synthesized. As illustrated in Fig. 1c and d, the result of flexible core
incorporation can be visualized, in an idealized two-dimensional
form, as creating a “checkerboard” of rigid (triazine) and more
flexible (propyl) network junctions in the cured state. In contrast,
networks formed from PT-30 exhibit junctions from either phenyl
rings or triazine rings, both of which are comparatively rigid. By
specifically designing the new monomer to closely resemble PT-30
in both equivalent weight, network junction density, and the ratio
of aliphatic to aromatic content, the effect of creating the “check-
erboard” of flexible network junctions was evaluated by compari-
sons to the properties of PT-30. Results showed that incorporation
of flexible junctions lowered the activation energy of cure, and
enabled nearly complete conversion to be achieved more readily.
The ability to achieve complete cure at a lower temperature can
improve performance by avoiding losses in properties that are
associated with cure at high temperatures [15]. These improve-
(1 L). The supernatant was decanted away from an insoluble
residue and then vigorously stirred while neutralized with HOAc.
Copious white solids precipitated. The precipitate was collected on
a coarse porosity glass frit and air-dried to constant weight. The
solid was recrystallized from EtOH to give the title compound as
colorless needles. Yield: 44 g (52%). Mp: 213e216 ꢀC (Ref. [16]
216e217 ꢀC). 1H NMR (DMSO-d6)
d (ppm): 12.31 (1H, CO2H), 7.68
(s, 1H), 7.09 (d, J ¼ 8.6 Hz, 2H), 7.04 (d, J ¼ 8.9 Hz, 2H), 6.95 (d,
J ¼ 8.7 Hz, 2H), 6.78 (d, J ¼ 8.7 Hz, 2H), 3.79 (s, 3H), 3.71 (s, 3H); 13C
NMR (DMSO-d6)
d (ppm): 168.75, 159.78, 158.54, 138.51, 131.84,
130.69, 130.38, 128.59, 127.01, 113.99, 113.80, 55.13, 55.01. Anal.
calcd for C17H16O4: C, 71.82; H, 5.67. Found: C, 71.75; H, 5.63. Trit-
uration of the insoluble residue with acetone gave 2 g of a white
solid. The solid was recrystallized from HOAc to give colorless flakes
found to be trans 4,40-dimethoxystilbene. Mp: 214e216 ꢀC (Ref. [17]
211e212 ꢀC). Anal. calcd for C16H16O2: C, 79.97; H, 6.71. Found: C,
78.60; H, 6.55.
2.2.2. Racemic 2,3-bis(4-methoxyphenyl)propionic acid (2)
A mixture of 1 (44 g, 0.15 mol), 5% Pd/C (500 mg) and DMF
(500 mL) was hydrogenated at rt and atmospheric pressure for
24 h. The mixture was filtered through diatomaceous earth to
remove the catalyst and the filtrate was poured into H2O (1 L). The
mixture was extracted three times with EtOAc (3 ꢁ 200 mL). The
extracts were collected and washed with H2O (300 mL) followed by
brine (300 mL). The organic layer was separated and dried over
anhydrous MgSO4. Rotary evaporation of the solvent left an off-
white solid. Recrystallization of the crude product from heptane/
EtOAc gave the title compound as colorless needles. Yield: 40.3 g
(90%). Mp: 125e127 ꢀC (Ref. [18] 125e127 ꢀC). 1H NMR (CDCl3)
d
(ppm): 11.47 (bs, CO2H), 7.19 (d, J ¼ 8.5 Hz, 2H), 6.99 (d, J ¼ 8.8 Hz,
2H), 6.82 (d, J ¼ 9.1 Hz, 2H), 6.73 (d, J ¼ 8.8 Hz, 2H), 3.78e3.69 (m,
7H), 3.29 (dd, J ¼ 13.8 and 8.1 Hz, 1H), 2.93 (dd, J ¼ 14.1 and 7.2 Hz,
1H); 13C NMR (CDCl3)
d (ppm): 180.04, 159.19, 158.33, 131.03,
130.26, 130.07, 129.34, 114.26, 113.97, 55.39, 55.33, 53.06, 38.65.
Anal. calcd for C17H18O4: C, 71.31; H, 6.34. Found: C, 71.39; H, 6.37.
ments were achieved while maintaining
a glass transition
temperature as high as the short-term thermal stability limit for
tricyanate esters, and with little or no loss in “wet” glass transition
temperatures. These favorable characteristics show that the incor-
poration of alternating flexible and rigid junctions in a rigid ther-
mosetting network is a highly useful, though mostly unexplored,
means of improving the properties of high-performance polymers.
2.2.3. Racemic 2,3-bis(4-methoxyphenyl)propionyl chloride (3)
A 1 L round-bottomed flask equipped with magnetic stirring bar
was charged with 2 (40.3 g, 0.14 mol) and SOCl2 (100 mL). A reflux
condenser with N2 bubbler piped to an aqueous NaOH gas scrub-
bing bottle was equipped and the mixture was gently refluxed.
After 3 h, the excess SOCl2 was evaporated and the crude oil was
dissolved in hexanes (300 mL). After stirring briefly, the product
precipitated as an off-white powder. The solid was collected by
filtration on a medium porosity glass frit. Recrystallization from
heptane gave the title compound as colorless needles. Yield: 33.1 g
2. Experimental
2.1. Materials
(77%). Mp: 62e65 ꢀC (Ref. [19] 64e66 ꢀC). 1H NMR (CDCl3)
d (ppm):
The 4-methoxyphenylacetic acid, p-anisaldehyde, acetic anhy-
dride (Ac2O), thionyl chloride (SOCl2), anisole, sodium borohydride
(NaBH4), 5% Pd/C catalyst, pyridine hydrochloride, cyanogen
bromide (BrCN), triethylamine (TEA) as well as solvents were
purchased from SigmaeAldrich (Milwaukee) and used as received.
7.16 (d, J ¼ 8.9 Hz, 2H), 6.98 (d, J ¼ 8.6 Hz, 2H), 6.87 (d, J ¼ 8.6 Hz,
2H), 6.76 (d, J ¼ 8.6 Hz, 2H), 4.16 (t, J ¼ 7.6 Hz, 1H), 3.77 (s, 3H), 3.74
(s, 3H), 3.39 (dd, J ¼ 14.2 and 7.6 Hz, 1H), 2.99 (dd, J ¼ 14.2 and
7.4 Hz, 1H); 13C NMR (CDCl3)
d (ppm): 174.75, 159.77, 158.62, 130.19,
129.79, 129.76, 127.58, 114.64, 114.09, 64.94, 55.42, 55.34, 38.86.
Anal. calcd for C17H17ClO3: C, 67.00; H, 5.62. Found: C, 67.12; H, 5.58.
2.2. Synthesis of monomer 9
2.2.4. Racemic 1,2,3-tris(4-methoxyphenyl)propanone (4)
2.2.1. (Z) 2,3-Bis(4-methoxyphenyl)acrylic acid (1)
A 500 mL round-bottomed flask equipped with magnetic stir-
ring bar was charged with 3 (30 g, 98 mmol) and anisole (21.1 g,
196 mmol, 2 equiv) and the mixture was gently heated for complete
dissolution. While vigorously stirred, AlCl3 (12.9 g, 98 mmol,
1 equiv) was added portionwise. Copious HCl gas was generated
during this time and the mixture became deep red in color. After all
the catalyst had been added, the mixture was heated to 70 ꢀC for
30 min. After this time, the mixture was cooled to rt and dissolved
in CH2Cl2 (300 mL). The reaction solution was slowly quenched by
A 1 L round-bottomed flask equipped with magnetic stirring bar
and reflux condenser charged with a mixture of 4-methox-
yphenylacetic acid (49.8 g, 0.3 mol), p-anisaldehyde (40.8 g,
0.3 mol, 1 equiv), TEA (30 mL, 21.78 g, 0.07 mol) and Ac2O (60 mL,
64.8 g, 0.21 mol) was refluxed for 4 h. The mixture was cooled to rt
and H2O (300 mL) was carefully added. The resulting pale yellow
precipitate was filtered on a coarse porosity glass frit. The crude
product was dissolved in a solution of KOH (16.8 g, 0.3 mol) in H2O

3936
A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
adding in a gentle stream to vigorously stirred H2O (300 mL). The
organic phase was separated and washed with H2O (300 mL) fol-
lowed by saturated aqueous NaHCO3 (300 mL) and finally brine
(300 mL). After drying the organic phase over anhydrous MgSO4,
the solvent and excess anisole were evaporated leaving a crude
white solid. Recrystallization from EtOH gave the title compound as
refluxed for 3 h by heating mantle after which time TLC showed the
reaction was complete. The mixture was cooled and carefully
quenched with H2O (500 mL). The mixture was extracted with Et2O
(3 ꢁ 200 mL). The extracts were collected and washed with 5% HCl
(300 mL) followed by H2O (500 mL) and finally brine (500 mL).
After drying over anhydrous MgSO4, the solvent was evaporated
leaving a thick colorless residue. Recrystallization from toluene
gave the title compound as a white microcrystalline powder. Yield:
22 g (94%). Mp: 161e163 ꢀC (Ref. [20,21] 164 ꢀC). 1H NMR (CDCl3/
a
white microcrystalline powder. Yield: 32.3
g
(87%). Mp:
(ppm): 7.89
116e119 ꢀC (Ref. [20,21] 122e124 ꢀC). 1H NMR (CDCl3)
d
(d, J ¼ 9.0 Hz, 2H), 7.13 (d, J ¼ 8.7 Hz, 2H), 6.98 (d, J ¼ 8.6 Hz, 2H),
6.82 (d, J ¼ 8.8 Hz, 2H), 6.78 (d, J ¼ 8.7 Hz, 2H), 6.73 (d, J ¼ 8.5 Hz,
2H), 4.66 (t, J ¼ 7.2 Hz, 1H), 3.79 (s, 3H), 3.74 (s, 3H), 3.73 (s, 3H),
3.44 (dd, J ¼ 13.7 and 7.3 Hz, 1H), 2.96 (dd, J ¼ 13.9 and 7.2 Hz, 1H);
DMSO-d6)
d (ppm): 8.19 (bs, 3OH), 6.86e6.75 (m, 6H), 6.65 (d,
J ¼ 8.9 Hz, 2H), 6.64 (d, J ¼ 8.3 Hz, 4H), 3.02 (m, 5H); 13C NMR
(CDCl3/DMSO-d6)
d (ppm): 154.74, 154.68, 135.03, 131.22, 129.64,
13C NMR (CDCl3)
d
(ppm): 198.37, 163.43, 158.79, 158.11, 132.36,
128.44, 114.77, 114.73, 48.95, 41.41, 20.69. Anal. calcd for C21H20O3:
C, 78.73; H, 6.29. Found: C, 78.59; H, 6.34.
131.86, 131.16, 130.29, 130.05, 129.49, 114.45, 113.86, 113.83, 55.59,
55.39, 55.06, 39.50. Anal. calcd for C24H24O4: C, 76.57; H, 6.43.
Found: C, 76.55; H, 6.43.
2.2.9. 1,2,3-Tris(4-cyanatophenyl)propane (9)
A 500 mL round-bottomed flask equipped with magnetic stir-
ring bar and 100 mL addition funnel was charged with 8 (26 g,
81 mmol), BrCN (30.1 g, 284 mmol, 3.5 equiv) and anhydrous
acetone (500 mL). The mixture was cooled in a ꢂ20 ꢀC bath and TEA
(24.6 g, 33.8 mL, 244 mmol, 3 equiv) was added dropwise over
30 min. Copious solids (TEA$HBr) precipitated during the addition.
The mixture was stirred in the cooling bath for 2 h and then
allowed to warm to rt. The reaction was quenched by slowly adding
H2O (400 mL) whereby the by-product salt dissolved and the
product precipitated concominantly. The white solid was collected
on a coarse porosity glass frit and air-dried under suction for
several hours. The crude product was recrystallized from i-PrOH to
give the title compound as colorless needles. Yield: 25.12 g (78%).
2.2.5. Racemic 1,2,3-tris(4-methoxyphenyl)propanol (5)
A 2 L round-bottomed flask equipped with magnetic stirring bar
and reflux condenser was charged with 4 (32.3 g, 85 mmol) and
anhydrous EtOH (500 mL). In several portions, NaBH4 (4.6 g,
121 mmol, 1.4 equiv) was added and the mixture was heated to
reflux. After 3 h, the reaction was complete by TLC. The mixture was
carefully diluted with H2O (500 mL) and stirred for 1 h. The mixture
was extracted Et2O (3 ꢁ 300 mL). The organic phases were collected
and washed with H2O (300 mL) followed by 300 mL brine. After
drying over anhydrous MgSO4, the organic layer was rotary evap-
orated leaving the title compound as a viscous, colorless oil. Yield:
30.68 g (95%). 1H NMR (CDCl3)
d (ppm):0.23e6.99 (m, 4H),
6.92e6.60 (m, 8H), 4.81 (d, J ¼ 6.3 Hz, 25% 1H), 4.74 (d, J ¼ 7.5 Hz,
Mp: 102e104 ꢀC 1H NMR (CDCl3)
7.09e6.99 (m, 6H), 3.18e3.00 (m, 3H), 2.97e2.85 (m, 2H); 13C NMR
(CDCl3) (ppm): 151.55,151.49,141.64,138.41,130.96,129.81,115.51,
d (ppm): 7.19e7.11 (m, 6H),
75% 1H), 3.87e3.62 (m, 9H), 3.26e2.57 (m, 3H), 1.83 (bs, OH); 13C
NMR (CDCl3)
d
(ppm): 159.38, 158.66, 157.87, 135.21, 132.52, 132.49,
d
130.24, 130.17, 130.05, 128.24, 127.98, 114.00, 113.92, 113.68, 113.59,
113.54, 55.88, 55.47, 55.37, 55.32, 55.28, 38.12. Anal. calcd for
C24H26O4: C, 76.17; H, 6.92. Found: C, 76.37; H, 7.04.
115.38, 108.90, 108.81, 49.50, 41.83. Anal. calcd for C24H17N3O3: C,
72.90; H, 4.33; N, 10.63. Found: C, 72.67; H, 4.25; N, 10.59.
2.3. Characterization
2.2.6. (E/Z) 1,2,3-Tris(4-methoxyphenyl)propene (6)
A 1 L round-bottomed flask equipped with magnetic stirring bar
and reflux condenser with DeaneStark attachment was charged
with 5 (30.68 g, 81 mmol), toluene (500 mL) and p-toluenesulfonic
acid dihydrate (380 mg). The mixture was brought to reflux and after
1 h the dehydration was complete. After cooling to rt, the mixture
was washed with saturated aqueous NaHCO3 (300 mL) followed by
brine (300 mL). After drying over anhydrous MgSO4, the solvent was
rotary evaporated leaving the title compound as a viscous, colorless
2.3.1. Monomer characteristics
Melting points were collected on a Mel-Temp II from Laboratory
Devices (Holliston, MA) and are not corrected. Infrared analysis of
monomer 9 and all cured and uncured samples, were analyzed by
Attenuated Total Reflection Fourier Transform Infrared (ATR-FTIR)
spectroscopy using a single bounce diamond ATR crystal. The
instrument used was a Nexus 870 FT-IR spectrometer with a liquid
N2 cooled mercury cadmium telluride (MCTA) detector. Each
spectrum is an average of 28 scans at 4 cmꢂ1 resolution. 1H NMR
and 13C NMR were collected on a Bruker Avance II 300 MHz spec-
trometer operated at 300 and 75 MHz respectively. Nuclear
magnetic resonance data (free-induction decays) were processed
using NUTS software from Acorn NMR (Livermore, CA). All spectra
are referenced to solvent or tetramethylsilane. Elemental analyses
were performed by Atlantic Microlab, Inc. (Norcross, GA).
CCDC 802927 (9) contains the supplementary crystallographic
data for this paper. These data can be obtained free of charge from
The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.
uk/data_request/cif, by e-mailing data_request@ccdc.cam.ac.uk, or
by contacting CCDC 12 Union Road, Cambridge CB2 1EZ, UK;
fax: þ44 1223 336033.
oil. Yield: 27.7 g (95%). 1H NMR (CDCl3)
d (ppm): 7.39e6.23 (m,13H),
3.96 (s, 60% 2H), 3.74e3.64 (m, 9H), 3.59 (s, 40% 2H); Anal. calcd for
C24H24O3: C, 79.97; H, 6.71. Found: C, 80.23; H, 6.76.
2.2.7. 1,2,3-Tris(4-methoxyphenyl)propane (7)
A mixture of 6 (27.7 g, 77 mmol), 5% Pd/C (1 g) and EtOH
(500 mL) was hydrogenated (2 psi) for 18 h. The mixture was
filtered through diatomaceous earth and rotary evaporated leaving
the crude product. Reduced pressure distillation (0.1 torr) gave the
title compound as a viscous, colorless oil. Yield: 26.5 g (95%). 1H
NMR (CDCl3)
d
(ppm): 6.89e6.78 (m, 6H), 6.69e6.61 (m, 6H), 3.68
(ppm):
(s, 3H), 3.67 (s, 6H), 2.98e2.64 (m, 5H); 13C NMR (CDCl3)
d
157.97, 157.87, 136.73, 132.96, 130.24, 129.00, 113.65 (2 overlapping
signals), 55.37, 55.34, 49.57, 41.88. Anal. calcd for C24H26O3: C,
79.53; H, 7.23. Found: C, 79.69; H, 7.26.
2.3.2. Cure kinetics
Studies of cure kinetics were performed via differential scanning
calorimetry (DSC) using w5 mg samples of uncatalyzed monomers
and a TA Instruments Q200 Differential Scanning Calorimeter
under 50 mL/min. of flowing nitrogen. The samples were first
heated to 120 ꢀC at 5 ꢀC/min. in order to produce molten material,
2.2.8. 1,2,3-Tris(4-hydroxyphenyl)propane (8)
A 500 mL round-bottomed flask equipped with magnetic stir-
ring bar and reflux condenser was charged with 7 (26.5 g, 73 mmol)
and pyridine hydrochloride (102 g, 12 equiv). The mixture was

A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
3937
then held at 120 ꢀC for 5 min to equilibrate and establish a baseline.
The samples were then heated as rapidly as possible (maximum
rate of around 100 ꢀC/min.) to the intended cure temperature, and
held at that temperature for 30 min. Following the hold at the cure
temperature, the samples were cooled as rapidly as possible (ca.
100 ꢀC/min.) to 120 ꢀC, and then re-heated to 350 ꢀC at 10 ꢀC/min. to
determine the residual enthalpy of polymerization. Cure temper-
atures of 210 ꢀCe290 ꢀC in increments of 20 ꢀC were used.
d
a
¼ k1 þ k2am
(4)
dt
ð1 ꢂ
n
a
Þ
in which a plot of the left-hand side of Eq. (4) as a function of yields
an intercept of k1 and a slope of k2. When, as expected, the values of
m and n did not vary significantly with temperature, the values of
k0 and k02 closely approximated the values of k1 and k2. The use of m
a1nd n was necessitated by the presence of significant systematic
correlations among the fitted values of k02, m, and n that could have
potentially biased the calculation of activation energies. Additional
details describing the collection (Section S2) and analysis (Section
S3) of the kinetic data are provided in the Supplementary Content.
To analyze the cure data, both the rate of conversion (d
a
/dt) and
the total conversion (a) were determined as follows:
t
Z
0
0
_
Qðt Þdt
_
d
a
ðtÞ
QðtÞ
t0
¼
r
a
ðtÞ ¼
(1)
2.3.3. Cured resin sample preparation and characterization
dt
QT þ QR
QT þ QR
Cured resin samples were prepared by either melting the
monomer powder, or thinning the monomer liquid in the case of
PrimasetÒ PT-30, at 120 ꢀC under a reduced pressure of 300 mm Hg
for 30 min, followed by pouring into either a reinforced silicone
_
where Q represents the baseline corrected differential heat flow, QT
is the total integrated heat flow (enthalpy of cure) observed during
the isothermal cure period beginning at t0, and QR is the sum of all
residual heat flow associated with resin cure observed on subse-
quent thermal cycling to 350 ꢀC. Since the cure of uncatalyzed
cyanate esters is expected to be auto-catalytic and to follow an n-th
order model with non-integer exponents (in contrast to catalyzed
systems that show second-order kinetics) [22], the cure data were
casting mold (either 13 mm diameter
ꢁ
3 mm discs, or
87 mm ꢁ 12 mm ꢁ 2.5 mm bars) for multi-step cure, or into a 6 mm
diameter ꢁ 1 mm high aluminum pan for single-step cure. Multi-
step cure involved heating in an oven under nitrogen to 150 ꢀC for
1 h, followed by 210 ꢀC for 24 h, with subsequent cooling to below
150 ꢀC prior to de-molding. Heating ramps were 5 ꢀC/min. Some
multi-step cured samples were then post-cured as free-standing
parts at temperatures of up to 290 ꢀC. The single-step cured
samples were heated in a small furnace under a nitrogen flow of
50 mL/min. with the temperature controlled to within 2 ꢀC, at
temperatures of 210 ꢀCe290 ꢀC, using the maximum possible ramp
rates (typically 50e100 ꢀC/min.). Additional details such as the mold
fabrication procedures have been published elsewhere [25].
Thermomechanical analysis of cured samples was performed
using a TA Instruments Q400 thermomechanical analyzer (TMA) in
dynamic (oscillatory compression) TMA mode. The oscillatory
compression mode provides qualitative information on the storage
and loss components of the sample stiffness, in addition to sample
displacement as a function of temperature. Due to the very high glass
transition temperatures of the samples studied and their tendency
to post-cure during heating to high temperatures, very rapid heating
and cooling rates of 50 ꢀC/minwere used. A standard thermal cycling
procedure, using limits of 0 ꢀC and 200 ꢀC, was used to determine
and correct for the thermal lag caused by these rapid heating rates.
The procedure was employed prior to heating to 400 ꢀC for dry
samples, and after heating to 350 ꢀC for wet samples. The mean
compressive load on the samples was 0.1 N, with an oscillatory force
applied at an amplitude of 0.1 N and a frequency of 0.05 Hz.
Complete details, examples, and the rationale for the cycling
procedures have been explained at length elsewhere [24]. Ther-
mogravimetric analysis (TGA) was carried out using a TA Instrument
Q5000 under 60 mL/min. of either nitrogen or air. For TGA analysis,
w2 mg of uncured powder was heated at 10 ꢀC/min. to 600 ꢀC. The
density of cured samples was determined via neutral buoyancy of
13 mm ꢁ 3 mm discs in CaCl2/deionized water mixtures, as
described elsewhere [26]. Some multi-step cured and post-cured
samples were placed in 250 mL of deionized water at 85 ꢀC for
96 h as a means of testing the effects of exposure to hot water.
fitted over the range
diffusion-controlled cure) to the Kamal model [23], namely:
a
¼ 0.1 to 0.5 (prior to the onset of significant
d
a
n
n
¼ k01ð1 ꢂ
a
Þ þk02amð1 ꢂ
a
Þ
(2)
dt
To fit the data, a modified form of the graphical analysis developed
by Kenny [24] was used. The method took advantage of the fact that
for the data collected, k01 << k20 , therefore the conversion at which
the rate of cure was maximum (amax) was very well approximated
by the quantity m/(m þ n). Since amax was readily observed at
multiple temperatures and was independent of temperature for
a given material, its value was used to constrain the fitting process.
A preliminary value of the catalytic rate constant, k01, was first
determined via extrapolation of the raw data per Kenny’s method.
Next, a trial value for n was chosen, and, in accordance with the
following rearrangement of Eq. (2) using preliminary rate constants
2
3
d
a
6
4
7
ꢂ k01 ¼ ln k02 þ m ln
a
(3)
dt
ln
5
ð1 ꢂ an
Þ
the quantity of the left-hand side of Eq. (3) was plotted as a function
of ln a
, enabling the computation of m and k02 via linear regression.
The value of m was then iteratively adjusted via the New-
toneRaphson method until the computed value of n and the chosen
value of m satisfied the constraint amax ¼ m/(m þ n). Due to the
special features of the particular data set being analyzed, this
modified method allowed values of all of the Kamal model
parameters to be determined through the solution of an objective
equation of the form f(d
a
/dt,
a
,
amax, k01, k02, m, n) ¼ 0 with the
straightforward ability to determine amax replacing the need to
judge the straightness of lines formed by plots of transformed data.
For the data under consideration, iterative solution of the objective
equation proved to be highly robust, and was therefore judged to be
a more appropriate method than the more generally applicable
graphical method described by Kenny.
3. Results and discussion
To determine the activation energies, values of m and n were
first determined separately for all suitable data sets describing
a single material. The mean values were then utilized to calculate
refined values of the rate constants k1 and k2 using a linear
regression of the Kamal model in the following form
3.1. Synthesis of tricyanate monomer 9
The synthesis of tricyanate monomer 9 is shown in Scheme 1.
Perkin condensation of 4-methoxyphenylacetic acid on anisalde-
hyde gave the 2,3-diarylacrylic acid 1 in modest yield [27e30].

3938
A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
Scheme 1. Synthesis of monomer 9: (a) 4-MeOC6H4CHO, Ac2O, TEA, reflux; (b) H2, 5% Pd/C, DMF; (c) SOCl2, heat; (d) anisole, AlCl3; (e) NaBH4, EtOH; (f) pTsOH, PhMe, reflux; (g) H2,
5% Pd/C, EtOH; (h) pyridine HCl, reflux; (i) BrCN, TEA, acetone, ꢂ20 ꢀC.
Catalytic hydrogenation of 1 in dimethylformamide solvent gave
the propionic acid 2 in excellent yield. Following acid chloride
generation, standard FriedeleCrafts reaction of 3 with anisole gave
the racemic ketone 4 previously reported by Kiprianov and Kut-
senko in good yield [20,21]. Reduction of 4 with sodium borohy-
dride gave a diastereomeric mixture of alcohol 5. Acid-catalyzed
dehydration of 5 gave a cis/trans mixture of 1,2,3-triarylpropene
6 which was then hydrogenated to give 7. The penultimate step
was demethylation of 7 with molten pyridine hydrochloride which
gave the solid triphenol 8 in excellent yield. Finally, reaction of 8
with cyanogen bromide in the presence of triethylamine gave the
solid tricyanate 9 in good yield. The 1H and 13C NMR of 9 is shown in
Fig. 2. The monomer 9 could be readily recrystallized from alcoholic
solvents and a crystal structure was obtained (provided in Section
S1 of Supplementary Content).
activation energy at elevated temperatures also appears to be
insensitive to impurities, and to decrease only moderately when
transition metal catalysts are present [34,35]. Thus, the enthalpy
and activation energy of cure were the focus of our analysis.
A sense of the overall cure kinetics of monomer 9 in comparison
to PrimasetÒ PT-30, a commercially available novolac-type cyanate
ester resin, is provided by the non-isothermal DSC thermograms
shown in Fig. 3. In both cases, the cure exotherm peaked at over
300 ꢀC, typical for a high purity cyanate ester monomer [36,37]. The
enthalpies of cure were also similar, at 750 J/g (99 kJ/eq.) for
monomer 9 and 660 J/g for PT-30. The enthalpy of cure for mono-
mer 9 on a cyanate ester equivalent basis was well within the
expected range [4,22] while that of PT-30 agreed well with previ-
ously reported values [38]. Based on multiple DSC samples heated
past the melting point, monomer 9 exhibited a melting endotherm
with a peak temperature of 107.4 ꢃ 0.2 ꢀC and an enthalpy of
melting of 83 ꢃ 1 J/g. No glass transition temperatures below 350 ꢀC
(at which point thermal degradation begins to interfere with DSC
analysis) were observed in the second heating scans of either
sample.
More detailed data on the kinetics of cure was provided by the
isothermal DSC data summarized in Fig. 4a (for monomer 9) and
b (for PrimasetÒ PT-30). A visual comparison of the two Figure-
s clearly reveals a more modest temperature dependence of the
kinetics of cure for monomer 9, indicative of a lower activation
energy. Though somewhat complex, the mathematical analysis of
the kinetic data is merely an elaboration on the preceding obser-
vation for the sake of quantification, with care being taken not to
bias the results through the choices inherent in the analysis
method. A detailed discussion may be found in Section S3 of the
Supplementary Content.
3.2. Cure kinetics of monomer 9
When discussing the cure kinetics of cyanate esters, it is espe-
cially important to take into consideration the effects of catalytic
impurities. Truly pure cyanate ester resins may not cure at all below
their chemical decomposition temperatures [31], and the overall
rates of cure are known to depend very strongly on the level of
phenolic and other impurities [22,32]. The NMR, elemental, and
X-ray analyses of monomer 9 were all indicative of impurities on
the order of 1% at most, and although no specific impurities were
identifiable, based on previous syntheses of cyanate esters [4,33],
partially cyanated phenolic compounds would be expected to
constitute the majority of any impurities present. Although
a detailed purity analysis could provide additional insights into
expected cure kinetics, it is much simpler to focus on aspects of
cure that are typically affected only modestly by impurities. For
cyanate esters, the total enthalpy of cyclotrimerization per equiv-
alent appears to be constant (within experimental error) [4] for
systems not intentionally catalyzed, and to show a very modest
decrease in systems catalyzed by transition metals [25]. The
The results of the kinetic analysis are summarized by the
Arrhenius plot of Fig. 5, which is based on the auto-catalytic rate
constant (k2) with temperature-independent reaction order expo-
nents. The activation energy for monomer 9 was modestly lower, at
110 ꢃ 2 kJ/mol, than for PT-30 (124 ꢃ 7 kJ/mol). The activation

A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
3939
Fig. 2. 1H and 13C NMR for monomer 9.
energies of cyanate ester cure vary widely [22], and the choice of
kinetic model can affect the values substantially, nonetheless these
values appear higher than those reported for more flexible dicya-
nate cyanate esters measured using similar techniques and
confirmed via multiple methods [39]. Importantly, the overall rates
of cure were similar for both materials through most of the
temperature range, implying the absence of any significant differ-
ences in catalytic impurity levels that could affect the activation
energy. At temperatures approaching 300 ꢀC, the PT-30 did cure at
a substantially faster rate, as would be expected based on the non-
isothermal DSC data shown in Fig. 3.
Fig. 4. Isothermal conversion history of (a) monomer 9 and (b) PT-30 as a function of
cure temperature.
Fig. 3. Non-isothermal DSC of monomer 9 and PrimasetÒ PT-30.

3940
A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
Although the PT-30 cured more rapidly than monomer 9 at
elevated temperatures, FT-IR analysis of the monomers, as high-
lighted in Fig. 6a (monomer 9) and b (PT-30), reveals that for single-
step cure at 290 ꢀC, monomer 9 achieved a higher absolute extent of
cure. Although these simple types of FT-IR comparisons cannot be
considered strictly quantitative, the observed difference in the
observed peak intensities near 2250 cmꢂ1 (indicative of residual
cyanate ester groups) were substantial enough to clearly show that
a lower overall degree of cure was achieved in PT-30 despite more
rapid kinetics. The ability of monomer 9 to achieve a greater extent
of cure was most likely due to the introduction of more flexible
network junctions, which allowed greater molecular motion even
after a significant cross-link density was produced, and in turn
enabled the residual cyanate ester groups to more easily rearrange
themselves to accommodate cyclotrimerization. Such decreased
rigidity at a molecular level is expected to affect the thermo-
mechanical behavior of the resin, which, given the high glass
transition temperatures of fully cured tricyanate esters, is likely to
be closely coupled to thermochemical behavior.
Fig. 5. Autocatalytic cure activation energy of monomer 9 and PT-30.
3.3. Physical properties
Fig. 7 shows oscillatory TMA data on both cured 9 and PT-30
samples. The TMA data shown in these Figures compares samples
after cure for 24 h at 210 ꢀC followed by 30 min of post-cure at
290 ꢀC. Such a multi-step cure was needed in order to prepare
Fig. 7. TMA of monomer 9 (thick blue lines) and PT-30 (thin red lines) after cure at
210 ꢀC for 24 h followed by 30 min at 290 ꢀC (a) before and (b) after 96 h immersion in
water at 85 ꢀC. Note that (b) shows data collected during the cooling ramp due to
bubble formation during heating. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
Fig. 6. FT-IR spectrum of (a) monomer 9 and (b) PT-30 before and after single-step
cure at 290 ꢀC for 30 min.

A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
3941
Table 1
Physical data for monomer 9 and PT-30 cured 24 h at 210 ꢀC followed by 30 min of
post-cure at 290 ꢀC.
Property
Cured 9
PrimasetÒ PT-30a
Density (g/cc)
1.251 ꢃ 0.001
1.260 ꢃ 0.001
Molar volume (cc/mol)
Cross-link density
at full cure (N/cc)b
316.0 ꢃ 0.4
302.7 ꢃ 0.3
3.811 ꢃ 0.004 ꢁ 1021
3.980 ꢃ 0.004 ꢁ 1021
a
Assumed structure as shown in Fig. 1 with n ¼ 1.
b
Includes central branch point as a cross-link.
Fig. 7b shows TMA scans for samples exposed to 85 ꢀC water for
96 h (the cooling scan is shown because it is unaffected by insta-
bilities related to the rapid drying of the sample during heating; the
heating scan is provided in Section S4 of Supplementary Content
and shows similar glass transition temperature amid a much
noisier signal). Both cured 9 and PrimasetÒ PT-30 exhibited glass
transition temperatures around 245 ꢀC. The weight gain on expo-
sure for identically cured samples was about 3.5% for cured 9
compared to 5.1% for PrimasetÒ PT-30. In addition to lower mois-
ture uptake, the more complete cure of monomer 9 provides the
benefit of greatly decreasing the formation of unstable carbamate
groups by the reaction of uncured cyanate ester groups with water
[40]. The improved relative thermomechanical performance of
cured 9 under wet conditions could not only result from better
hydrolytic stability and lower moisture uptake, it could also result
partly from the fact the mechanical softening alone, rather than
a combination of mechanical and chemical changes, determined
the glass transition temperature under these conditions. These data
indicate that for many applications, in which performance under
wet conditions is a key design consideration, there will be little or
no difference in maximum allowable use temperature between
cured 9 and PrimasetÒ PT-30.
Fig. 8 shows the TGA curves under both nitrogen and air for
cured 9 (Fig. 8a) and PrimasetÒ PT-30 (Fig. 8b). In accord with the
TMA results described earlier, cured 9 showed decreased thermal
stability under both nitrogen and air compared to PT-30, with the
onset of degradation about 30 ꢀC lower, and about twice as much
mass loss. These results make sense given the presence of less
thermally stable substituted propyl groups in monomer 9. In the
case of thermochemical stability, the enhanced molecular motion
afforded by the presence of these groups may also have facilitated
chemical degradation. Thus, decreased thermochemical stability
was one of the primary trade-offs associated with the introduction
of flexible junctions in the network. It should be noted that the
thermochemical stability of cured 9 was nonetheless quite good in
comparison with many types of thermosetting polymers, including
dicyanate esters, with a char yield near 50% at 600 ꢀC in air. Addi-
tional physical properties, such as density and nominal cross-link
density, are shown in Table 1.
Fig. 8. TGA of (a) monomer 9 and (b) PT-30 under nitrogen and air.
uncatalyzed cyanate ester samples of sufficient size and quality for
mechanical analysis, because a single jump to 290 ꢀC resulted in the
production of a significant quantity of volatiles (most likely due to
an appreciable vapor pressure of the monomer itself at this
temperature). By achieving the maximum extent of cure practically
attainable at 210 ꢀC, both the generation of volatiles as well as
previously observed thermochemical degradation of the cyanate
ester network during the subsequent post-cure was minimized.
Note also that thermochemical degradation of the samples began to
take place before the completion of the scan, thus the drop in
stiffness was likely due to both physical and chemical changes in
the sample.
As seen in Fig. 7a, the dry glass transition temperatures for both
samples are very high, at least 340 ꢀC for cured 9 and at least 390 ꢀC
for PrimasetÒ PT-30. Thus, the data in Fig. 7 are in agreement with
previously observed [11] values for the glass transition temperature
of fully cured PT-30. The earlier termination of the TMA run was
due to an apparent decrease in the thermochemical stability of the
cured 9. In the absence of chemical degradation, the softening point
of a typical thermosetting network would likely be lowered by the
introduction of more flexible network junctions, however, in this
particular case, the TMA experiment revealed that the maximum
use temperature under dry conditions was determined by chemical
stability considerations. Thus, for dry samples of cured 9 the
increased flexibility did not impact maximum use temperatures
compared to PT-30 by sufficiently lowering the mechanical soft-
ening temperature of the chemically in-tact network, but rather by
decreasing the chemical stability of the network at elevated
temperatures.
4. Conclusions
A new tricyanate ester monomer featuring flexible chemical
linkages at the central branch point [1,2,3-tris(4-cyanatophenyl)
propane] has exhibited a number of useful performance improve-
ments over a more rigid analog (PrimasetÒ PT-30), demonstrating
the potential value of introducing a “checkerboard” architecture of
alternating rigid and flexible junctions between molecular
segments of macromolecular networks. In particular, the new
monomer, which was synthesized in nine steps with an overall
yield of 26%, exhibited an activation energy of 110 kJ/mol for auto-
catalytic cure at temperatures of 210 ꢀCe290 ꢀC, 14 kJ/mol lower
than the activation energy of PrimasetÒ PT-30 under identical
conditions. The overall extent of cure achievable at temperatures of

3942
A.J. Guenthner et al. / Polymer 52 (2011) 3933e3942
290 ꢀC was also higher for the new monomer. In terms of physical
properties, the new monomer showed a dry glass transition
temperature of at least 340 ꢀC, likely limited by a somewhat lower
thermochemical stability compared to PrimasetÒ PT-30, while both
materials exhibited a wet glass transition temperature around
245 ꢀC. These results indicate that many of limitations associated
with rigid thermosetting monomers having a high cross-link
density may be overcome with only modest trade-offs through
a carefully chosen introduction of flexible junctions in the molec-
ular architecture of the network.
[10] Ganguli S, Dean D, Jordan K, Price G, Vaia R. Polymer 2003;44(22):6901e11.
[11] Marella V. Masters Thesis. Drexel University; 2008.
[12] Ganguli S, Dean D, Jordan K, Price G, Vaia R. Polymer 2003;44(4):1315e9.
[13] Yameen B, Duran H, Best A, Jonas U, Steinhart M, Knoll W. Macromol Chem
Phys 2008;209(16):1673e85.
[14] Cambrea LR, Davis MC, Groshens TJ, Guenthner AJ, Lamison KR, Mabry JM.
J Polym Sci Polym Chem 2010;48(20):4547e54.
[15] Goertzen WK, Kessler MR. Compos Appl Sci Manuf 2007;38(3):779e84.
[16] Ketcham R, Jambotka D. J Org Chem 1963;28(4):1034e7.
[17] Spatz SM. J Org Chem 1961;26(10):4158e61.
[18] Meyers MJ, Sun J, Carlson KE, Marriner GA, Katzenellenbogen BS,
Katzenellenbogen JA. J Med Chem 2001;44(24):4230e51.
[19] Kölling H, Lettré H. US Patent 2,691,044; 1954.
[20] Kiprianov GI, Kutsenko LM. Ukr Khim Zh 1957;23:505e9.
[21] Kiprianov GI, Kutsenko LM. Chem Abstr 1958;52:6280a.
[22] Hamerton I, Emsley AM, Howlin BJ, Klewpatinond P, Takeda S. Polymer 2003;
44(17):4839e52.
Acknowledgments
[23] Kamal MR, Sourour S. Polym Eng Sci 1973;13(1):59e64.
[24] Kenny JM. J Appl Polym Sci 1994;51(4):761e4.
Financial support through an In-house Laboratory Independent
Research award from the Office of Naval Research, as well as
funding from the Air Force Office of Scientific Research, is gratefully
acknowledged.
[25] Guenthner AJ, Yandek GR, Mabry JM, Lamison KR, Vij V, Davis MC, Cambrea
LR. Insights into moisture uptake and processability from new cyanate ester
monomer and blend studies. In: SAMPE International Technical Conference,
vol. 55. Salt Lake City, UT: SAMPE International Business Office, 2010. p.
42ISTC-119.
[26] Lamison KR, Guenthner AJ, Vij V, Mabry JM. Packing fraction and relation to
glass transition in ternary blends of cyanate ester resins. PMSE Prepr 2011;
104.
Appendix. Supplementary material
Supplementary data related to this article can be found online at
doi:10.1016/j.polymer.2011.07.024.
[27] Buckles RE, Bremer K. Org Synth 1953;33:70e2.
[28] Oxley P, Short WF. GB Patent 559,024; 1944.
[29] Green FD, Adam W, Cantrill JE. J Am Chem Soc 1961;83(16):3461e8.
[30] Ruan B, Yang Y, Zhu Z, Lu P, Zhu H. Acta Crystallogr Sect E Struct Rep Online
2009;65:o944.
References
[31] Bauer M. Thesis. Academy of Sciences of the GDR; 1980.
[32] Georjon O, Galy J, Pascault JP. J Appl Polym Sci 1993;49(8):1441e52.
[33] Guenthner AJ, Yandek GR, Wright ME, Petteys BJ, Quintana R, Connor D, et al.
Macromolecules 2006;39(18):6046e53.
[34] Yan HQ, Ji L, Qi GR. J Appl Polym Sci 2004;91(6):3927e39.
[35] Recalde IB, Recalde D, García-Lopera R, Gómez CM. Eur Polym J 2005;41(11):
2635e43.
[36] Laskoski M, Dominguez DD, Keller TM. J Polym Sci Polym Chem 2006;44(15):
4559e65.
[37] Zhao L, Hu XA. Polymer 2010;51(16):3814e20.
[1] Lin CH, Yang KZ, Leu TS, Sie JW. J Polym Sci Polym Chem 2006;44(11):
3487e502.
[2] Papathomas KI. US Patent 5,292,861; 1994.
[3] Jackson RJ. US Patent 4,916,210; 1990.
[4] Hamerton I. Chemistry and technology of cyanate ester resins. London:
Chapman & Hall; 1994.
[5] Fang T, Shimp DA. Prog Polym Sci 1995;20(1):61e118.
[6] Nair CPR, Mathew D, Ninan KN. Cyanate ester resins, recent developments.
New Polymerization techniques and Synthetic Methodologies. Adv Polym Sci
2001;155:1e99.
[7] McConnell VP. Resins for the hot zone part II: BMIs, CEs, benzoxazines &
phthalonitriles. High Perform Compos; 2009:43e9.
[8] Wienhold PD, Persons DF. SAMPE J 2003;39(6):6e17.
[9] Reams JT, Boyles DA. J Appl Polym Sci 2011;121(2):756e63.
[38] Chen CC, Don TM, Lin TH, Cheng LP. J Appl Polym Sci 2004;92(5):3067e79.
[39] Sheng X, Akinc M, Kessler MR. J Therm Anal Calorim 2008;93(1):77e85.
[40] Shimp DA, Ising SJ. Moisture effects and their control in the curing of poly-
cyanate resins. PMSE Prepr 1992;66:504e5.